Month: April 2014

Look into the past, become more agile: egoless programming

Agile software development methods like extreme programming and Scrum have been around for almost 20 years now. Many think of them as “the right way” or some kind of “holy grail”. So it is no wonder that many companies try to hop on the agile train to improve quality and efficiency of their software development.

Why is it that I hear of failures in adopting/switching to agile most of the time?

In my experience there are two main reasons why implementing agile methods does not work as expected:

  1. It is the management deciding to change the development process, not the developers. After a short while many aspects – especially using rigid methods like Scrum – do not seem to fit the organisation and therefore are changed. This leads to process implementations like ScrumBut
  2. Many developers do not have the mindset for the interactive, open and collaborative work style required by agile methods. Too often developers try to secure their job by isolating themselves and their solutions. Or they are doing stuff the way they are used to for many years not willing to learn, change and improve. Communication is hard, especially for programming nerds…

What can be done about it?

I would suggest trying to slowly change the culture in the company based on concepts from the 1970’s: egoless programming. In essence you have to let developers become more open and collaborative by internalising the Ten Commandments of Egoless Programming:

  1. Understand and accept that you will make mistakes.
  2. You are not your code.
  3. No matter how much “karate” you know, someone else will always know more.
  4. Don’t rewrite code without consultation.
  5. Treat people who know less than you with respect, deference, and patience.
  6. The only constant in the world is change.
  7. The only true authority stems from knowledge, not from position.
  8. Fight for what you believe, but gracefully accept defeat.
  9. Don’t be “the guy in the room.”
  10. Critique code instead of people—be kind to the coder, not to the code.

There is a PDF version of the commandments with some amplifying sentenctes from Builder.com published on techrepublic in 2002. If you manage to create a development culture based on these values you are more than half-way down the agile road and can try to implement one of the popular agile methods – or your own!

Developer power tools: Big O = big impact

Scalability and performance are often used interchangeable but they are very different. The big O notation helps in talking about scalability.

Scalability and performance are often used interchangeable but they are very different. The resource consumption of a program is its performance. So performance looks at a specific part of a program, say the user activating an action and tries to fix the input and circumstances.
Scalability defines how the resource consumption of a program changes when additional resources are added or the input size increases. So a program can have good performance (it runs fast) but when the load increases it can behavely badly (a bad scalability). Take bubble sort for example. Since the algorithm has low overhead it runs fast with small arrays but when the arrays get bigger the runtime gets worse. It doesn’t scale.
There are many ways to improve scalability here we look at one particular: reducing algorithmic complexity. The time complexity of an algorithm is measured with the big T notation. The T notation describes the time behavior of the algorithm when the input size changes. Say you have an algorithm that takes n objects and needs 4 times as long with an overhead of 2. This would be:

T(n) = 4n + 2

Often the correct numbers are not necessary, you just want to have a magnitude. This is where the big O notation comes into play. It describes the asymptotical upper bound or just the behaviour of the fastest growing part. In the above case this would be:

T(n) = 4n + 2 = O(n)

We call such an algorithm linear because the resource consumption grows linear with the increase in input size. Why does this matter? Let’s look at an example.

Say we have a list of numbers and want to know the highest. How long would this take? A straight forward implementation would look like this:

int max = Integer.MIN_VALUE;
for (int number : list) {
  if (number > max) {
    max = number;
  }
}

So we need if the size of the number list is n we need n compare operations. So our algorithm is linear. What if we have two lists of length n? We just our algorithm twice and compare both maximums so it would be

T(n) = 2n + 1 = O(n)

also linear.
Why does this all matter? If our algorithm runs quadratic, so O(n^2) and our input size is n = 1000, we need a multitude of 1 million operations. If it is linear it just needs a multitude of 1000 operations, much less. Reducing the complexity of an algorithm can be business critical or the difference between getting an instant result or losing a customer because he waited too long.
Time complexity isn’t the only complexity you can measure other resources like space can also be measured with big O. I hope this clears some questions and if you have further ones please feel free and leave a comment.

Gigapixel images in pure Java

What you can do when you hit an ancient limitation of Java while working with gigapixel sized images.

Not long ago, I read this nice little blog entry about the basic properties and usages of Java arrays. It’s a long time since I last used an array in Java myself, because my programming style evolved to heavily leverage the power of collections (and Iterables in particular, the Java 5 poor man’s substitute for Java 8 Streams). But I immediately noticed that one important fact was missing from the array blog entry:

The maximum length of an array in Java is Integer.MAX_VALUE or ((2^32)-1), aka 2.147.483.647

This is indirectly specified in the Java language specification, chapter 10.4 Array Access:

Arrays must be indexed by int values.

This little fact crossed my path when writing a little tool in pure Java that operated on large numbers of large images, combining them to a gigantic image. The customer used the tool to create images that had a size of about 100 MB, but took several hours to print because the decompression tax kicked in. One day, he reported a strange bug:

array-error-cropped

 

“Oh, a negative array size, what a strange bug to appear in a tested application” was my first thought. Only after reading the stacktrace more carefully did it dawn on me: The array size wasn’t negative, it was just bigger than Integer.MAX_VALUE and got wrapped around into the negative numbers. And sure enough, 72350 times 44914 is a respectable 3.249.527.900 pixels, more than 1,5 times as much as an array in Java can hold. This image was right in the multi-gigapixel range where all kinds of technical obstacles appear. The maximum length of an array in Java was mine.

Trying to stay pure

One cornerstone of the tool was being lightweight. It shouldn’t carry around unnecessary luggage and weighted around 200 kB when the bug appeared – enough to just copy it into the data directories instead of pulling the directories into the program. But when I examined the root cause of the problem at hand, I found the frustrating truth that Java’s built-in imaging library also relies on one cornerstone: all data is stored in one array. And this array can only hold around 2G entries of data.

My approach was to “partition” the full image into smaller parts that only stored a fraction of the overall pixels. To hide this fact from the ImageIO that ultimatively writes all the data into one file, my PartitionedImage implements RenderedImage and has to translate every call into a series of appropriate subcalls to the partition images. Before we look at some code, let me show you the limitations of this approach:

Greedy JPEGs, credulous PNGs

In the RenderedImage interface, there are two methods that can be used to obtain pixel data:

  • Raster getData(): Returns the image as one large tile (for tile based images this will require fetching the whole image and copying the image data over).
  • Raster getData(Rectangle rect): Computes and returns an arbitrary region of the RenderedImage.

If an image writer calls the first method, my code is screwed. There is no mentally sane way to construct a Raster instance without colliding with the array length limitation. Unfortunately, the JPEG writer does just that: He gets greedy and demands all the pixels at once. I found it easier to avoid the JPEG format and therefore trade disk space for pragmatism.

The PNG writer uses the getData(Rectangle) method to obtain the pixel data. It calls the whole image line by line: the region has always the full width of the image, but is only one pixel in height. So I guess my tool will write a lot of large PNG images in the future.

Our partitions should adapt to this behaviour by always retaining the full width of the original image and only allowing enough height that the amount of pixels per partition doesn’t exceed Integer.MAX_VALUE.

The remaining trick is to implement an AdjustingRaster that knows the original Raster of the partition and translates the row asked by the PNG writer to the according row in the partition image. The AdjustingRaster needs to know about the vertical offset. The only pitfall here is that the vertical offset has to be zero while the AdjustingRaster gets written to and needs to be set once it switches into read mode.

Slow, but working

By composing a gigapixel image from several partitions (sometimes called tiles) you can circumnavigate the frustrating limitation of Java’s arrays (I mean, it’s 2014 and 64-bit systems are somewhat prevailing now. No need to stick to 32-bit limits without a good reason). The result isn’t overwhelmingly fast, but I suspect that’s caused by the PNG image writer more than by our indirections. And we shouldn’t forget that it’s a lot of pixels to write after all.

Conclusion

Sometimes when you explore bigger and bigger use cases, you hit some arbitrary limitation. And some are fundamental ones. In our case here, we’ve reached the limit of Java arrays and got stuck because the image library in Java never heard of real gigapixel imaging and coupled itself hard to the array limit. By introducing another indirection layer on top of the image library implementation and using composition to emulate a bigger image than we actually could create, we can convince non-sceptical image writers to save all those pixels for us and even manipulate the image beforehand.

What was your approach for gigapixel image processing? How did it work out in the long run? Share your story in the comments, please.

Dart and TypeScript as JavaScript alternatives

JavaScript was designed at Netscape by Brendan Eich within a couple of weeks as a simple scripting language for the web browser. It’s an interesting mixture of Self‘s prototype-based object model, first-class functions inspired by LISP, a C/AWK-like syntax and a misleading name imposed by marketing.

Unfortunately, the haste in which JavaScript was designed by a single person shows in many places. Lots of features are inconsistent and violate the principle of least surprise. Just skim through the JavaScript Garden to get an idea.

Another aspect casting a poor light on JavaScript is the bad design of the browser DOM API, including incompatibilities between different browser implementations.

Douglas Crockford redeemed the reputation of JavaScript somewhat, by writing articles like “JavaScript: The World’s Most Misunderstood Programming Language“, the (relatively thin) book “JavaScript: The Good Parts” and discovering the JSON format. But even his book consists for the most part of advice on how to avoid the bad and the ugly parts.

However, JavaScript is ubiquitous. It is the world’s most widely deployed programming language, it’s the only programming language option available in all browsers on all platforms. The browser DOM API incompatibilities were ironed out by libraries like jQuery. And thanks to the JavaScript engine performance race started by Google some time ago with their V8 engine, there are now implementations available with decent performance – at least for a scripting language.

Some people even started to like JavaScript and are writing server-side code in it, for example the node.js community. People write office suites, emulators and 3D games in JavaScript. Atwood’s Law seems to be confirmed: “Any application that can be written in JavaScript, will eventually be written in JavaScript.”

Trans-compiling to JavaScript is a huge thing. There are countless transpilers of existing or new programming languages to JavaScript. One of these, CoffeeScript, is a syntactic sugar mixture of Ruby and Python on top of JavaScript semantics, and has gained some name recognition and adoption, at least in the Rails community.

But there are two other JavaScript alternatives, backed by large companies, which also happen to be browser manufacturers: Dart by Google and TypeScript by Microsoft. Both have recently reached version 1.0 (Dart even 1.2), and I will have a look at them in this blog post.

Large-scale application development and types

Scripting languages with dynamic type systems are neat and flexible for small and medium sized projects, but there is evidence that organizations with large code bases and large teams prefer at least some amount of static types. For example, Google developed the Google Web Toolkit, which compiled Java to JavaScript and the Closure compiler, which adds type information and checks to JavaScript via special comments, and now Dart. Facebook recently announced their Hack language, which adds a static type system to PHP, and Microsoft develops TypeScript, a static type add-on to JavaScript.

The reasoning is that additional type information can help finding bugs earlier, improve tool support, e.g. auto-completion in IDEs and refactoring capabilities such as safe, project-wide renaming of identifiers. Types can also help VMs with performance optimization.

TypeScript

This weekend the release of TypeScript 1.0 was announced by Microsoft’s language designer Anders Hejlsberg, designer of C#, also known as the creator of the Turbo Pascal compiler and Delphi.

TypeScript is not a completely new language. It’s a superset of JavaScript that mainly adds optional type information to the language via Pascal-like colon notation. Every JavaScript program is also a valid TypeScript program.

The TypeScript compiler tsc takes .ts files and translates them into .js files. The output code does not change a lot and is almost the same code that you would write by hand in JavaScript, but with erased type annotations. It does not add any runtime overhead.

The type system is heavily based on type inference. The compiler tries to infer as much type information as possible by following the flow of types through the code.

TypeScript has interfaces that are very similar to interfaces in Go: A type does not have to declare which interfaces it implements. Interfaces are satisfied implicitly if a type has all the required methods and properties – in short, TypeScript has a structural type system.

Type definitions for existing APIs and libraries such as the browser DOM API, jQuery, AngularJS, Underscore.js, etc. can be added via .d.ts files.
These definition files are very similar to C header files and contain type signatures of the API’s functions. There’s a community maintained repository of .d.ts files called Definitely Typed for almost all popular JavaScript libraries.

TypeScript also enhances JavaScript with functionaliy that is planned for ECMAScript 6, such as classes, inheritance, modules and shorthand lambda expressions. The syntax is the same as the proposed ES6 syntax, and the generated code follows the usual JavaScript patterns.

TypeScript is an open source project under Apache License 2.0. The project even accepts contributions and pull-requests (yes, Microsoft). Microsoft has integrated TypeScript support into Visual Studio 2013, but there is support for other IDEs and editors such as JetBrain’s IDEA or Sublime Text.

Dart

Dart is a JavaScript alternative developed by Google. Two of the main brains behind Dart are Lars Bak and Gilad Bracha. In the early 90s they worked in the Self VM group at Sun. Then they left Sun for LongView Technologies (Animorphic Systems), a company that developed Strongtalk, a statically typed variant of Smalltalk, and later the now-famous HotSpot VM for Java. Sun bought LongView Technologies and made HotSpot Java’s default VM. Bracha co-authored parts of the Java specification, and designed an object-oriented language in the tradition of Self and Smalltalk called Newspeak. At Google, Lars Bak was head developer of the V8 JavaScript engine team.

Unlike TypeScript, Dart is not a JavaScript superset, but a language of its own. It’s a curly-braces-and-semicolons language that aims for familiarity. The object model is very similar to Java: it has classes, inheritance, abstract classes and methods, and an @override annotation. But it also has the usual grab bag of features that “more sugar than Java but similar” languages like C#, Groovy or JetBrain’s Kotlin have:

Lambdas (via the fat arrow =>), mixins, operator overloading, properties (uniform access for getters and setters), string interpolation, multi-line strings (in triple quotes), collection literals, map access via [], default values for arguments, optional arguments.

Like TypeScript, Dart allows optional type annotations. Wrong type annotations do not stop Dart programs from executing, but they produce warnings. It has a simple notion of generics, which are optional as well.

Everything in Dart is an object and every variable can be nullable. There are no visibility modifiers like public or private: identifiers starting with an underscore are private. The “truthiness” rules are simple compared to JavaScript: all values except true are false.

Dart comes with batteries included: it has a standard library offering collections, APIs for asynchronous programming (event streams, futures), a sane HTML/DOM API, removing the need for jQuery, unit testing and support for interoperating with JavaScript. A port of Angular.js to Dart exists as well and is called AngularDart.

Dart supports a CSP-like concurrency model based on isolates – independent worker threads that don’t share memory and can communicate via SendPorts and
ReceivePorts.

However, the Dart language is only one half of the Dart project. The other important half is the Dart VM. Dart can be compiled to JavaScript for compatibility with every browser, but it offers enhanced performance compared to JavaScript when the code is directly executed on the Dart VM.

Dart is an open source project under BSD license. Google provides an Eclipse based IDE for Dart called the “Dart Editor” and Dartium, a special build of the Chromium browser that includes the Dart VM.

Conclusion

TypeScript follows a less radical approach than Dart. It’s a typed superset of JavaScript and existing JavaScript projects can be converted to TypeScript simply by renaming the source files from *.js to *.ts. Type annotations can be added gradually. It would even be simple to switch back from TypeScript to JavaScript, because the generated JavaScript code is extremely close to the original source code.

Dart is a more ambitious project. It comes with a new VM and offers performance improvements. It will be interesting to see if Google is going to ship Chrome with the Dart VM one day.